Study Design

Important considerations for the design of studies to develop a prognostic model include the advantages of a prospective study, the need for patients to be followed-up from a common event (such as surgery or diagnosis), and the quality of measurements. In prospective studies, eligible patients are enrolled, complete baseline measurements are made in a standardized way, and they are followed for an adequate length of time to allow a comparison of survival experience in relation to baseline variables including tumor marker values. Some randomized trials sensibly incorporate the collection of tumor markers for prognostic studies; otherwise, prospective studies are rare. Most prognostic studies in cancer use existing clinical data, however. These studies have the considerable advantage of the ready availability of a cohort with a long enough follow-up for assessment of a substantial number of outcome events (e.g., deaths or recurrences). But the cohort may not be complete, measurements may not be standardized, and some baseline data may be missing. Also, treatments given to patients will usually vary, with the choice of treatment partly related to prognostic information.

0 0

Post a comment